In multi-level data packet-switched systems, demodulators are often used to accomplish data detection. The filters used to band-limit the modulated IF spectrum are typically not symmetric about the IF center frequency. Since the output of demodulators are AC coupled, the coupling capacitor used will charge to the noise idle or average value of the noise being provided by the filter. This average noise voltage may be offset somewhat from the demodulator voltage produced by a on-channel signal, so that when the receiver goes from a no signal condition to receiving a packet, there is a long RC time constant due to the fact that a large capacitor must be used to pass the low frequency components of the data signal. This time constant is sufficiently long so as to appear as a DC offset when observing in the time frame of a single packet.
The problem caused by the DC offsets is that, in typical multi-level data detection schemes, an absolute DC level is used to establish differing "zones" for detection. If the average noise voltage (DC offset) is high, demodulation errors will occur. The DC offsets that are present at the output of the demodulator can be manually adjusted to compensate for the offset, but eventually the preset manual adjustments will drift over time due to changing temperature and the corresponding change in the component characteristics. Other implementations have used a feed-forward system in trying to send a carrier before the detector or demodulator begins to detect information. This method however, consumes a fair amount of space in the packet, reducing the amount of available data throughput.
Thus, a need exists for a DC offset correction system that continuously corrects for DC offset while not requiring the use of a broadcasted carrier.